Emotions & TV Viewership
Massive Twitter Study Proves Correlation of Emotions and TV Viewing

Overview
Ask anybody who’s ever watched and loved a TV show why they continue to tune in, and they’ll most likely respond with an emotion. “I love Jon Snow,” or “the action scenes are crazy!” or “the plot twists are so intense!” Does this mean that emotions drive viewership?
In order to test the correlation between TV viewership and emotions, Canvs conducted the most extensive study of Twitter TV data ever. We classified the Emotional Reactions to over 5,700 TV episodes across numerous program genres using Canvs’ patent-pending technology. And then attempted to predict how viewership would increase or decrease week to week based upon changes in Emotional Reactions.
As the industry standard in emotion measurement, Canvs is uniquely qualified to conduct this study. Our Canvs TV solution is proven to be 200% more accurate than traditional sentiment analysis at analyzing how Millennials feel. Canvs is unmatched in understanding the language of social conversations. Our knowledge base includes over 5 million idioms, slang expressions, acronyms, and misspellings not found in the dictionary, but commonly found in social.

Methodology:
5,709 episodes of 431 TV series airing between January 1, 2014 - June 30, 2015
Adult 18-49 Viewership in Millions (Live+Same Day)

Control for:
Episode number
Program genre: Reality, Comedy, Drama
Total number of episode tweets

Predict using:
Canvs Emotion Rates (e.g. %Love, %Crazy, etc.)
Viewership source: Nielsen as reported by tvbythenumbers.com; Program Genre: Sourced from Gracenote; Tweet source: Twitter data from Nielsen, which measures program-related Twitter activity for linear episode airings and on a 24/7 basis; Episode number: Finales are flagged; Sample: Broadcast made up 40%; Cable made up 55%; Premium made up 5%. 29% of episodes were in first season.
Key Findings
Based on Canvs’ measurement of the emotions expressed on Twitter, we have developed a well-calibrated probability that viewership will go up or down next week. The key emotions that drive viewership vary across comedy, reality, and drama genres, and vary in level of importance across the genres. The key emotions, their predictive power by genre, and sample Tweets are below:

COMEDY
Read as: In Comedy, for every 1% increase in Beautiful, there is a .3% increase in viewership.
Beautiful ( +.3%)
Love (+.1%)

REALITY
Read as: In Reality, for every 1% increase in hate, there is a .7% increase in viewership.
Hate (+.7%)
Crazy ( +.3%)
Love (+.2%)

DRAMA
Read as: In Drama, for every 1% increase in hate, there is a .7% increase in viewership.
Hate (+.7%)
Funny (+.3%)
Love (+.3%)
The Canvs TV Viewership Study proves that Emotional Reactions expressed on social media can predict program tune-in. Canvs has leveraged the viewership study to create the Canvs Viewership Probability (or CVP), which uses a well-calibrated, statistically-derived percentage that viewership will go up or down for a reality, drama, or comedy TV show, available nearly one week in advance of next week’s airing. To find out more, please complete the form below or watch this recent clip from Fox 5 NY.
Additional resources
Variety Data Science Proves We Love to Hate-Watch TV
AdWeek A Study of TV-Related Tweets Finds Hate More Than Love Drives Viewership
Multichannel News Study: 'Hate’ Drives Viewers to Return to Dramas
Broadcasting & Cable Study: 'Hate’ Drives Viewers to Return to Dramas
The Wrap Hate Drives TV Viewers to Return to Shows, New Study Finds
The Drum Canvs study ties emotional reactions in TV-related tweets to viewership changes